field ion microscopy of platinum adatoms deposited on a...

*Corresponding author. Tel.: #49-30-8413-4180; fax: #49-30-8413-4101.

Ultramicroscopy 79 (1999) 231}238

Field ion microscopy of platinum adatoms depositedon a thin Al

2O

3"lm on NiAl(1 1 0)

N. Ernst*, B. Duncombe, G. Bozdech, M. Naschitzki, H.-J. Freund

Department of Chemical Physics, Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4}6, D-14195 Berlin, Germany

Received 12 September 1998; received in revised form 14 February 1999

Abstract

Aiming to contribute to a characterisation of the surface di!usion and clustering of individual metal adatoms ona well-characterised oxide "lm the feasibility of FIM investigations is demonstrated for the catalytically relevant systemPt/Al

2O

3/NiAl(1 1 0). A thin aluminium oxide "lm was grown on a [1 1 0] oriented NiAl "eld emitter through

dissociative oxygen adsorption followed by heating cycles. The "lm was characterised using FEM, FES and neon FIM.Field desorption sequences revealed that the "lm prepared on NiAl(1 1 0) was approximately two layers thick. Below80 K the "lm was stable at "eld-strengths applied for neon FIM. The presence of an ordered structure was suggested byFEM and FIM pattern. FIM of Pt adatoms was achieved at 79 K and below best image "eld strength using neon as theimaging gas. From preliminary displacement observations the surface di!usion activation energy of individual Ptadatoms was estimated and ranges between 0.4 and 0.5 eV. After three heating cycles to a maximum temperature of230 K an ordered arrangement of Pt adatoms was noticed in the vicinity of a presumable point defect. ( 1999 ElsevierScience B.V. All rights reserved.

Keywords: Thin oxide "lm; Individual, transition-metal adatoms; Surface di!usion

1. Introduction

Thin oxide "lms, supporting small metal aggreg-ates, are of great technological importance sincethey can act as model systems for heterogeneouscatalysis [1,2]. Following a relatively early report[3] on the structural characterisation of an orderedaluminium oxide on NiAl(1 1 0) a series of invest-igations was carried out on this system using thearsenal of modern surface science techniques in-cluding high-resolution imaging methods, such asHREM and STM. Charging e!ects limit the ap-

plication of most surface physics techniques toextremely thin "lms which can be produced indi!erent ways. When the surface of a NiAl crystal isexposed to oxygen the formation of an alumina "lmis thermodynamically strongly favoured at properconditions. Structural and electronic properties ofsuch aluminium oxide "lms have been established,especially for the closed packed NiAl(1 1 0) surface.In this case a well ordered and approximately0.5 nm thin aluminium oxide "lm can be grown [2].The various experimental results obtained suggestthat the stoichiometry and the structure of theoxide is compatible with c-Al

2O

3. A schematic

structure model of the oxide "lm is sketched inFig. 1.

0304-3991/99/$ - see front matter ( 1999 Elsevier Science B.V. All rights reserved.PII: S 0 3 0 4 - 3 9 9 1 ( 9 9 ) 0 0 0 8 9 - 3

Fig. 1. Schematic structure model of the aluminium oxide "lm on NiAl(1 1 0) [2]. Unit meshes of domains A and B of the thin oxide "lmare shown. The rectangle in the middle illustrates the symmetry of the NiAl(1 1 0) substrate surface and the distorted hexagon indicatesthe arrangement of oxygen ions in domain B of the alumina "lm.

In the "eld emission community attempts on thepreparation and characterisation of metals on lessordered and relatively thick oxides, such asNi/Rh

2O

3[4] and Pd-Mo/Al

2O

3/W have been

reported [5]. Using an atom probe FIM the oxida-tion of a Ni-rich NiAl alloy has been characterised.However, no ordering of the oxide "lm was re-ported [6]. We are not aware of any reports onFIM, HREM and STM observations of individualmetal adatoms deposited on a well-ordered oxide.Aiming at the study of the energetics and kinetics ofthe di!usion and clustering of individual metaladatoms on an ordered oxide "lm we have appliedFIM and "eld electron spectroscopic techniques onthe system Al

2O

3on NiAl(1 1 0). The main goal of

the present work has been: (i) preparation andcharacterisation of an ordered Al

2O

3"lm on

a [1 1 0] oriented NiAl tip and (ii) FIM of Ptadatoms deposited on the alumina "lm.

2. Experimental procedures

The thin oxide "lm has been prepared accordingto previously described procedures [7]. The NiAltips used as substrates for the oxidation processwere produced in four steps: (i) Cutting of [1 1 0]oriented rods from a single crystal, (ii) cleaning of

the rods through heating in UHV, (iii) electrolyticetching in air and "nally (iv) preparation of the tipin an FIM set-up through heating cycles (to a max-imum temperature of 1300 K), neon-ion sputteringand "eld evaporation. Pulsed FEV at ¹"50 K hasbeen employed which should reduce the e!ect ofa preferential removal of aluminium ions [8] de-teriorating the formation of an alumina "lm duringoxidation. For preparation of the aluminium oxidelayer the cleanliness of the NiAl tips was checked,e.g. through a "nal FEM picture before dosing thetip with O

2. At the initial stage of the experiments

the dose was 1500 L at 350 K guided by earliermeasurements [7,9]. After measurements of oxygeninduced work function changes on NiAl(1 1 0), weincreased the doses in order to secure a completeO coverage at 350 K. An ordered alumina "lm wasformed through a number of thermal #ashes (toa maximum temperature of 1300 K) and after eachsingle #ash an electron current measurement wastaken accompanied by an FEM picture. When theelectron current remained constant the oxide "lmwas imaged in neon. Eventually, the "lm was "eldevaporated producing a clean NiAl-substrate sur-face followed by a new preparation sequence.

Two FIM/FEM set ups have been used. The "rstone allowed liquid-nitrogen cooling of the speci-men [10] and has been equipped with a platinum

232 N. Ernst et al. / Ultramicroscopy 79 (1999) 231}238

Fig. 2. (a) Neon FIM of [1 1 0] oriented NiAl prepared by pulsed FEV at 50 K. (b) After #ashing to 1300 K the diameter of the (1 1 0)plane has increased by one order of magnitude to approx. 50 nm. (c) After oxidation followed by FD an aluminium oxide island is left onthe NiAl(1 1 0) apex plane. (d) The FEM pattern shows CsCl (bcc-like) symmetry and the boundary of the oxide island. The dead spothas been produced by an inactive area in the channel plate.

evaporator. It was built using a heatable tungstenloop which was covered with a small piece of plati-num wire. After several outgassing cycles the Pt#ux was calibrated using a tungsten tip. The vac-uum was improved through an additional sublima-tion pump so that a base pressures below10~11 mbar was reached. After exploratory FIMon Pt/Al

2O

3/NiAl(1 1 0) at 79 K a new specimen

holder was installed allowing gaseous or liquidhelium cooling of the specimen. FIM at 50 K wasthus made possible leading to signi"cantly im-proved image qualities. For probe hole FIM/FEMmeasurements a second set-up allowed "eld-ionand electron spectroscopy of selected surface sites[11]. Energy analyses were carried out in a retard-ing potential and a hemispherical mirror analyser.The latter was especially used for measurements ofthe low energy side of electron energy distributions.

From energy distribution and i}< (Fowler}Nord-heim) measurements work function data ofNiAl(1 1 0) were determined [12].

3. Results and discussion

To convey an impression of the results of thepreparation procedures we present observations ofthe clean and of the oxidised NiAl tip. The imagesshown in Fig. 2 have been obtained at ¹(NiAl)"50 K allowing neon FIM at good resolution andcontrast conditions. Pulsed FEV has been em-ployed for cleaning the specimen surface usinga pulse to DC fraction of more than 20% but notexceeding 40%. In the example shown in Fig. 2(a)a pulse voltage of 3.2 kV was superimposedon a DC voltage of 8 kV in order to minimise the

N. Ernst et al. / Ultramicroscopy 79 (1999) 231}238 233

Fig. 3. Probe}hole FEM analyses of clean [1 1 0] oriented NiAl. (a) Two Fowler}Nordheim (FN) plots and (b) one total energydistribution are shown. From the slopes, m

FN(1 1 0) and m

ED(1 1 0) work function values of NiAl(1 1 0) are obtained. Values for m

FN(total)

were derived from measurements of the total current through the tip.

e!ect of preferential FEV of aluminium ions.One can also restore the surface concentration of Alby #ashes to 1300 K as will further be discussedbelow. Our observations show that #ashing to thistemperature does not blunt the tip. We have routine-ly used this technique to approach the thermal endform. Fig. 2(b) shows the shape of a thermallyprepared tip characterised by an enlarged (1 1 0)plane. Compared to a low-temperature FEV shapea thermally enlarged NiAl(1 1 0) facet can of coursebe covered by a larger number of Al

2O

3unit cells.

The oxide "lm was formed applying a procedurewhich included dissociative O

2adsorption and

#ash heating to temperatures ranging between700 K and 1300 K [7]. FIM observations during"eld desorption revealed that the oxide "lm cover-ing the (1 1 0) facet was approximately two layersthick. It was stable in high "eld conditions neededfor low temperature ((80 K) neon FIM(+33 V/nm). It could be `peeled o!a from the sur-face in a controlled manner through "eld desorp-tion. The layer on the NiAl(1 1 0) apex plane cameo! at the end and showed greatest stability. Oxideislands of a desired size as in Fig. 2(c) and (d) canroutinely be prepared now. The oxide showed some

amount of order as deduced from FIM and FEMobservations. The boundary between the oxide is-land and the NiAl substrate could clearly be vis-ualised even in the FEM pattern in Fig. 2(d). Inview of the experience assembled for metal substra-tes, such oxide islands should be well suited forFIM studies of the di!usion of adatoms such asplatinum.

To further characterise the clean and theoxidised NiAl(1 1 0) surface we have carried outprobe hole FEM analyses. Fig. 3 shows the resultsof i}< and total energy distribution measurementsfor electrons "eld emitted from the centre ofa NiAl(1 1 0) plane prepared by DC FEV at 79 K.Work function data derived from such measure-ments [12] for di!erent pretreatments are given inTable 1. A thermal #ash to 1300 K reduces the workfunction of those NiAl(1 1 0) surfaces which hadbeen prepared by DC FEV at 79 K. This changeagrees with earlier photo emission data on macro-scopic NiAl(1 1 0) samples correlating the decreas-ing work function with a shift in the stoichiometryto the ideal 50%Ni/50%Al ratio [13].

Fig. 4 shows "eld electron energy spectra derivedfrom retarding potential analyses of clean, O-covered


Fig. 4. (a) Retardation curves of "eld emitted electrons were measured for clean and oxidised NiAl(1 1 0) and for the surface coveredwith a saturation layer of chemisorbed oxygen. (b) Electron energy distributions have been obtained by di!erentiation. At the Fermienergy the shift of the onsets for clean and oxidised NiAl(1 1 0) is within the experimental uncertainty (30 meV).

Table 1Work function data of clean NiAl

Pretreatment Work function (eV)

(1 1 0) Total

d.c. FEV, 79 K 5.6 4.5Flash, 1300 K 5.1 }

and oxidised NiAl(1 1 0). The spectra showed on-sets at the Fermi level which were identical within30 meV for the clean NiAl(1 1 0) and the oxidisedsurface in agreement with the result of an earlierFES measurement [14]. We conclude that the thinAl

2O

3"lm is highly transparent for electrons of the

NiAl substrate at the Fermi energy in an appliedelectric "eld of approx. 4 V/nm. It should benoted that the "eld emission current in Fig. 4(a)has decreased by approximately a factor of twoafter the oxide growth had been completed onNiAl(110). Exposing a thermally cleaned NiAl(1 1 0),as in Fig. 4(a) to 6000 L O

2at ¹(NiAl)"

350 K, lowers the local emission current by a factorof 30. This is interpreted as a signi"cant workfunction increase possibly caused by chemisorbedoxygen. A partial oxidation of the NiAl(1 1 0) sur-face can also not be excluded [9]. In conclusion, the

probe hole FEM approach gives valuable datawhich are particularly helpful for the proceduresinvolved in the preparation of the oxide "lm.

Finally, we report the results of our attempts toimage individual platin adatoms deposited on theoxide "lm at 79 K using neon FIM. The best image"eld strength was not applied since the "lm alreadystarted to "eld desorb at 79 K at that condition.This was veri"ed after the adatom imaging se-quence had been "nished. For these reasons neonFIM was characterised by strong #uctuations andnon-optimum contrast. These drawbacks couldpartly be compensated by computer aided super-position of selected video frames. The contrast re-versed pattern in Fig. 5(a) has been obtainedby superposing 44 single frames taken at 9.0 and9.5 kV, Neon-BIV+10.7 kV, FEV(NiAl)+12.6 kV,FD(Al

2O

3/NiAl(1 1 0))+10.6 kV at 79K. This proce-

dure al-lowed to visualise weak features, such as theones located on the boundary of the uppermostoxide layer. Four emission centres located on thetopmost Al

2O

3plane could be identi"ed. Since "eld

desorption of the oxide "lm even in peripheralregions of tip was avoided in this particular experi-ment, the origin of the four emission centres on thebare alumina substrate remains obscure. It is con-ceivable that these spots are images of locally pro-truding substrate atoms. Alternatively, one could


Fig. 5. Neon FIM of Pt/Al2O

3/NiAl(1 1 0) at 79 K. Time-averaged video frames displayed in negative contrast mode show the surface of

the aluminium oxide "lm: (a) Before and (b) after deposition of platinum. Images (a) and (b) were obtained from FIM observations atapplied voltages between 9.0 kV and 9.5 kV (Neon BIV+10.7 kV). The polygon curves indicate the boundary of the topmost Al

2O

3layer (diameter ca. 25 nm). (c) After three heating cycles of approx. 10 s at 200}230 K, neon FIM was done at higher applied voltage,9.9 kV (FD (Al

2O

3)+10.6 kV)). (d) A model has been outlined for positions of Pt adatoms locally arranged, close to a presumble defect

site (arrow).

presume the presence of point defects such as singleadatoms on an otherwise perfect Al

2O

3surface.

Intending to observe cluster formation of plati-num adatoms, the oxide surface was exposed toa relatively high platinum dose at "eld free condi-tions and at ¹(Al

2O

3)"79 K. The dose (o!ered for

3 min) had earlier been calibrated by means ofa [1 1 0] oriented tungsten tip and amounted toapprox. 15 Pt atoms per Al

2O

3unit cell, 1.9 nm2 as

sketched in Fig. 1. After Pt deposition the FIMpattern in Fig. 5(b) was obtained at the same ap-plied voltages as in Fig. 5(a). Thirty spots can becounted, some of which have already been identi-"ed at 8.8 kV. These spots are located within thepolygon curve indicating the boundary of the top-most oxide plane of the tip apex. Nine additional

spots located very close to the boundary have beendiscarded. The imaged objects represent mostprobably individual Pt adatoms. On the basisof earlier experiences [2] exchange with substrateatoms and interdi!usion processes through the"lm are excluded for present deposition conditions(¹(Al

2O

3)"79 K). The counted number of

spots (30) gives only a lower limit of the actualnumber of deposited Pt atoms since the appliedelectric "eld strength was not high enough toidentify all adatoms, at least at this stage ofthe experiment. Almost all spots observed afterPt deposition appear to provide images of indi-vidual adatoms. One exception is a possible dimercon"guration spotted at approx. ten o'clock inFig. 5(b).


1Recently, two of us (G.B. and N.E.) measured the onset temper-ature for the surface di!usion of three Pt adatoms deposited onAl

2O

3/NiAl(1 1 0) employing Neon-FIM at 45 K: ¹

0/4%5"165 K.

Using Eqs. (1) and (2) one obtains a value for the surface di!u-sion activation energy, E

$+0.45 eV (D

0"7.7]10~4 cm2/s,

¹"165 K, Sr2T"9]10~16 cm2, t"20 s, l"3]10~8 cm).

Table 2Estimation of surface di!usion parameter for Pt/Al

2O

3/

Ni(1 1 0)!

Displacement(nm)

Di!usivity(nm2/s)

Activation energy(eV)

0.8 0.014 0.5124 15 0.39

! ¹"200 K, t"10 s.

In spite of limitations governing the imagingprocess at 79 K, we have obtained "rst hints on thelocal ordering of Pt adatoms. This was visualised atsomewhat higher applied voltage, 9.9 kV which wasstill well below the desorption voltage for the oxide"lm, 10.6 kV. After three heating periods of approx.10 s at 200}230 K ordering occurred in the vicinityof a presumable defect site close to the step edge(Fig. 5(c)). We can not exclude that adatoms hadcrossed the step-boundary of the uppermost layer(in both directions), especially during the last heat-ing step when the surface temperature reached230 K. At this temperature (and above) we ob-served a signi"cantly changing total number ofadatoms on the topmost plane. For an examinationof the energetic properties of the step edges moresingle adatom experiments are desirable. Since thevariation of the magni"cation has not been mappedout a reliable distance calibration can not be givenat present. In view of our current experimentalanalyses we favour a calibration which would resultin a unit mesh distance (Fig. 5(d)) of approx. 0.6 nm.This implies a local arrangement of the Pt adatomscharacterised by a relatively open structure. It isstriking, however, that the alignment of theadatoms is compatible with directions given by thelong and the short axis of the Al

2O

3unit cell of

domain B (Fig. 1).To initiate surface di!usion of the Pt adatoms

the oxide was heated four times: (i) 200 K 10 s (ii)220 K 13 s (ii) 230 K 10 s and (iv) 240 K 10 s at "eldfree conditions. From the FIM observations madebefore and after (i) we conclude that the adatomsstart to jump already at 200 K (and possibly at aneven lower temperature1). From FIM images of theclean NiAl tip, taken before oxide preparation andafter removal by FD, we got a rough estimate of thediameter of the topmost oxide plane in Fig. 5 ofapprox. 25 nm. Using this calibration we deter-mined minimum and maximum values for adatomdisplacements allowing to estimate values of the

activation energy of surface di!usion. For an esti-mate of the (2-dim. random walk) di!usivity,

D"D0exp (!E

$/k¹)"Sr2T/4t, (1)

a value for the prefactor has been calculated, usingthe relation

D0+0.25l2k¹/h. (2)

Here, E$

is the activation energy of surface di!u-sion, k and h are Boltzmann's and Planck's con-stants. Minimum and maximum values for themean square displacement Sr2T were obtainedfrom present FIM observations. The elementarydi!usion length, l was assumed as 0.3 nm as sugges-ted by the surface structure of Al

2O

3/NiAl(1 1 0)

shown in Fig. 1. Numerical values of the estimatedsurface di!usion parameter are given in Table 2. Itis clear, however, that a precise experimental deter-mination of surface di!usion parameter requiresmore accurate measurements of the displacement,especially at di!erent surface temperatures.

4. Conclusion

It appears feasible that selected, relatively strong-ly bound adatom species, deposited on a thinAl

2O

3"lm on NiAl(1 1 0) can be imaged by low

temperature neon FIM as veri"ed for Pt in experi-ments at 79 K. From the present FIM observationswe conclude that Pt adatoms are quite mobile at200 K and start to jump at a signi"cantly lowertemperature. Experiments carried out below 50 Kshould improve the conditions for FIM ofPt/Al

2O

3/NiAl(1 1 0) thus allowing to characterise

surface di!usion properties of individual platinumadatoms in more detail.


Acknowledgements

We gratefully acknowledge the help of Drs. JoK rgUnger and Yuri Vlasov in setting up the probe holeFIM apparatus and Dr. Bernhard Sieben in hand-ling the computer aided image analyses. We wouldalso like to thank Professor Gert Ehrlich for helpfulcomments. One of us (N.E.) gratefully acknow-ledges travel support from the German ResearchFoundation (DFG).

References

[1] C.T. Campbell, Surf. Sci. Rep. 227 (1997) 1.[2] H.-J. Freund, Angew. Chem. Int. Ed. 36 (1997) 452.

[3] K. MuK ller, H. Lindner, D.M. Zehner, G. Ownby, Verhand-lungen der Deutschen Physikalischen Gesellschaft 25(1990) 1130.

[4] G.L. Kellogg, Surf. Sci. 257 (1991) 1.[5] J. Sotola, Z. Knor, J. Catal. 145 (1994) 501.[6] D.G. Ren, Surf. Rev. Lett. 2 (1995) 177.[7] R.M. Jaeger, H. Kuhlenbeck, H.-J. Freund, M. Wuttig, W.

Ho!mann, R. Franchy, H. Ibach, Surf. Sci. 259 (1991) 235.[8] M.K. Miller, R. Jayaram, Surf. Sci. 266 (1992) 458.[9] H. Isern, G.R. Castro, Surf. Sci. 211/212 (1986) 865.

[10] Y.-S Lim, M. Berdau, M. Naschitzki, M. Ehsasi, J.H.Block, J. Catal. 149 (1994) 292.

[11] J. Unger, Yu.A. Vlasov, N. Ernst, Appl. Surf. Sci. 87/88(1995) 106.

[12] A. Modinos, Field, Thermionic, and Secondary ElectronEmission Spectroscopy, Plenum Press, New York andLondon, 1984, p. 53.

[13] M. Onellion, J.L. Erskine, Phys. Rev. B 36 (1987) 4495.[14] K. Bobev, V. Gaidarova, Phys. Stat. Sol. A 23 (1974) K129.


field ion microscopy of platinum adatoms deposited on a...

Documents